Research On WSN Data Compression Based On Predictive Class Algorithm

Wireless sensor networks(WSN)consist of many micro sensor nodes in the monitoring environment.They cooperate with each other and transmit the collected signals to the control center through wireless communication and multi hop self-organization.WSN has the characteristics of rapid deployment,automatic networking,high concealment and high fault tolerance,but its computing and storage capabilities and power and energy are very limited.So,it's very important to study data compression methods in wireless sensor networks.How to reduce the computational complexity and energy consumption of sensor nodes while improving the accuracy and efficiency of data compression is one of the urgent problems to be solved in data compression of wireless sensor network.This paper mainly works on the WSN data compression technology,and research that how to reduce energy consumption and improve the efficiency of WSN network data compression.We introduce the research background and significance of the subject,give an overview of WSN data compression and WSN related knowledge.It also introduces several widely used data compression algorithms including the main ideas,advantages and disadvantages.In view of the high complexity of calculation,low compression efficiency and data recovery accuracy rate of situation,three solutions are given in this paper.First,a WSN data compression method based on spatial correlation and grey model(Piecewise Data Compression Method for Grey Model,GMPDC)is proposed.The method is based on the WSN monolayer cluster structure.First,the sensor nodes are required to send the collected data segments.Then the data space redundancy is eliminated by using the WSN data compression method with low computational complexity in the cluster head node.Finally,the compressed data is restored through the grey model in the base station.In addition,the application of grey model to WSN data compression is proved to be effective,through analysing the data recovery effect of the grey Markov chain model and the grey model under the condition of different length and compression rate.Finally,the simulation results show that the proposed cluster head and base station separated structure of method can significantly reduce the data transmission volume and improve the compression accuracy and efficiency.Secondly,a WSN data compression method based on improved grey model(Piecewise Data Compression Method for Odd and Even Grey Model,OEGMP)is proposed.Based on the GMPDC,the method considering the WSN data transmission mechanism to improve the grey model,making it more suitable for the requirements of reducing transport capacity and improve the accuracy of WSN data compression.According to the experiments,the data recovery effect of the improved grey model is better than the grey model and the grey Markov chain model at the same compression rate,and the optimal model and length of the algorithm are given according to the data recovery effect of the three models under the condition of different length and compression rate.The simulation results show that,compared with the GMPDC,the proposed improving method can effectively improve the compression accuracy and efficiency.Finally,different from eliminating the spatial redundancy of GMPDC and OEGMP,aiming at the correlation between multiple parameters of single node of WSN data,a WSN data compression method based on Hidden Markov model(Piecewise Data Compression Method for Hidden Markov Model,HMMPDC)is proposed.Based on the separate structure of the node base station of WSN,the method determines the benchmark data set and the optimal parameter of the model by analyzing the historical data at first.Then the benchmark data set and noncorrelation data are segmented by polynomial fitting,and some correlation data are sent,after Correlation grouping the parameters collected by a single node at intervals.Finally,the Hidden Markov Model(HMM)is used to recover the non sent part of the correlation data in the WSN base station.Simulation results show that compared with traditional data compression methods,the proposed method can effectively reduce network data transmission and has strong applicability and scalability.